# Which Claude Code mods should a beginner try first? Start with mods that only watch, then one that steps in. Five Claude Code mods for beginners, where they work, and a 60-second safety check. URL: https://ergostrat.com/en/automations/claude-code-mods-first.html Service: https://ergostrat.com/en/automations.html Checked: 2026-10-08 Contact: contact@ergostrat.com Language: English. Български: https://ergostrat.com/automations/claude-code-mods-first.html Start with mods that only watch, such as a gauge of how full your context window is, and move to a mod that steps in, such as one that stops risky commands, only once you know how to check what a mod can reach. Claude Code mods need version 2.1.287 or later and run in the terminal and in the Code tab of the Claude Desktop app, not in the claude.ai chat. Before you load any mod, one command lists what it listens to and what it asks Claude Code to do. ## What is a Claude Code mod? A mod is a plugin that changes how Claude Code looks and behaves, written as small functions that Claude Code calls when something happens. Anthropic's documentation describes a mod as JavaScript or TypeScript event handlers. When Claude is about to use a tool, when you send a prompt, or when part of the screen is drawn, the mod's function runs and can watch the event, change it, or take it over. That is what lets a mod draw a band above the prompt, add a command of your own, or hold a command until you say yes. You do not have to write the code yourself. In a normal Claude Code session you describe the mod in your own words, Claude writes it, and Claude Code asks whether to turn on hot reloading for the session. A mod made this way loads only in the session that made it, so to keep it you copy its folder somewhere of your own. ## Where do Claude Code mods work? In the terminal and in the Code tab of the Claude Desktop app, with Claude Code 2.1.287 or later in the terminal. Type claude --version in a terminal to see your version. The Desktop app carries its own copy of Claude Code, and the documentation says mods work there from 2.1.286. In the VS Code extension's chat panel and in scripted runs a mod's functions still run, but nothing it draws appears. Mods do not run in a Desktop session under WSL, and the claude.ai chat in a browser is a different product, so if that is the only place you use Claude, mods are not for you yet. ## Which Claude Code mods for beginners come first? Go from mods that only watch to mods that step in, so each one teaches you something before the next one can change anything. 1. Token Weather, one of Anthropic's three sample mods. It draws a short forecast of how full your context window is above the prompt and reads usage numbers only. 2. Replay Theater, also an Anthropic sample. It adds a replay command that steps through the file edits Claude made in the last reply, one change at a time. 3. A mod you ask Claude to write, for example one that pings you when a long reply ends. It only watches when replies end, so it is a safe way to learn the loop of describe, approve, try, ask for a change. 4. Blast Radius, the third Anthropic sample. It holds a risky shell command, such as deleting a folder or a force push, and shows what it would change, with buttons to proceed or cancel. Try it in an empty test folder first. 5. A usage gauge that shows your weekly and five-hour limits and your context window in one line above the prompt. Our own version, called Fuel gauge, comes with the guide described below. Anthropic shares the three samples in its claude-code-playground repository on GitHub, as they are and without support. To try one, load its folder for a single session with the plugin-dir option; close the session and it is gone. ## How do you check a mod before you load it? Run claude plugin validate on the mod's folder and read two lines of the output: hooks and calls. A mod is not sandboxed. Anthropic's documentation says plainly that once it loads it runs with your permissions: it can read and write your files, start programs, use the network, read environment variables where API keys often sit, see every prompt and tool call, and spend your usage. That does not make mods dangerous by default; it makes them like any program you install. The hooks line lists the events the mod listens to. The calls line lists what it asks Claude Code to do. A mod that only draws should need little beyond drawing calls. Words about files, processes, the network or environment variables each need a reason you can say out loud. If you cannot find one, do not load the mod, or ask Claude to read it and explain each call in plain words first. If something goes wrong, starting Claude Code with the safe-mode option turns off every mod you installed for that session. ## Is there a guide to Claude Code mods in plain English? ErgoStrat wrote one for people who use Claude Code but do not want to read developer documentation: Claude Code Mods in Plain English. The guide walks through the five mods above in the same order, with nine copy-paste prompts that make useful mods for you, the 60-second safety check, a fix-it list for a mod that does nothing, and the Fuel gauge mod ready to load. Each prompt was run in a fresh Claude Code 2.1.288 session and the mod it made passed claude plugin validate. Mods change between Claude Code releases, so the guide names the version it was tested on and is updated free for 90 days after purchase. It is sold as a PDF download on Etsy. It is an independent guide, not made or endorsed by Anthropic. ## Questions ### Do I need to know how to code to use Claude Code mods? No. You can describe a mod in a normal session and let Claude write it, or load one someone else wrote. You do need to approve what Claude writes and to run the safety check before you load a mod from someone you do not know. ### How do I turn all mods off? For one session, start Claude Code with the safe-mode option. For every session, set disableAllHooks to true in your Claude Code settings file; that also stops your settings hooks and your custom status line. ### Will a mod keep working after Claude Code updates? Not always. The events and methods a mod uses can change between releases, which is why Anthropic suggests saying in a mod's README which version it was tested with. ## Sources - [Anthropic, "Mods overview", Claude Code documentation, 2026.](https://code.claude.com/docs/en/plugins/mods/overview) - [Anthropic, "Create a mod", Claude Code documentation, 2026.](https://code.claude.com/docs/en/plugins/mods/create) - [Anthropic, "claude-code-playground: sample mods", GitHub, 2026.](https://github.com/anthropics/claude-code-playground/tree/main/claude-code/mods) - [ErgoStrat, "Claude Code Mods Guide in Plain English", Etsy, 2026.](https://www.etsy.com/listing/4588019344/claude-code-mods-guide-in-plain-english) ## More on this topic - [n8n vs Zapier vs Make: which one should you choose? Zapier is the easiest and has the most integrations. Make combines power with ease of use. n8n is the only one with self-hosting and prices in EUR.](https://ergostrat.com/en/automations/n8n-make-zapier.html) - [Can business process automation be done without code? No-code automation is built by dragging blocks, not by programming. Which tools allow it (Activepieces has over 450 integrations) and where they fall short.](https://ergostrat.com/en/automations/no-code-automation.html) - [What is the difference between RPA and business process automation? RPA copies clicks on a screen for one task. Business process automation connects a whole process through APIs. When each one earns its place.](https://ergostrat.com/en/automations/rpa-vs-process-automation.html) ## When the work outgrows one tool, it becomes an automation. A mod changes how Claude Code behaves for one person at one desk. When the same work runs across email, documents and orders, we map it and build the automation that carries it.